[Bug tree-optimization/43065] [4.5 Regression] gcc.c-torture/execute/20051215-1.c is miscompiled with -fgraphite-identity
--- Comment #10 from spop at gcc dot gnu dot org 2010-03-08 17:50 --- Subject: Bug 43065 Author: spop Date: Mon Mar 8 17:49:48 2010 New Revision: 157288 URL: http://gcc.gnu.org/viewcvs?root=gccview=revrev=157288 Log: Add testcase from PR43065. 2010-03-04 Sebastian Pop sebastian@amd.com PR middle-end/43065 * gcc.dg/graphite/run-id-3.c: New. Added: trunk/gcc/testsuite/gcc.dg/graphite/run-id-3.c Modified: trunk/gcc/ChangeLog.graphite -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=43065
[Bug tree-optimization/43065] [4.5 Regression] gcc.c-torture/execute/20051215-1.c is miscompiled with -fgraphite-identity
--- Comment #11 from spop at gcc dot gnu dot org 2010-03-08 17:50 --- Subject: Bug 43065 Author: spop Date: Mon Mar 8 17:49:57 2010 New Revision: 157289 URL: http://gcc.gnu.org/viewcvs?root=gccview=revrev=157289 Log: Fix PR43065: Insert bounds on pointer type parameters. 2010-03-05 Sebastian Pop sebastian@amd.com Reza Yazdani reza.yazd...@amd.com PR middle-end/43065 * graphite-sese-to-poly.c (add_param_constraints): Insert bounds on pointer type parameters. * gcc.dg/graphite/run-id-4.c: New. Added: trunk/gcc/testsuite/gcc.dg/graphite/run-id-4.c Modified: trunk/gcc/ChangeLog.graphite trunk/gcc/graphite-sese-to-poly.c -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=43065
[Bug tree-optimization/43065] [4.5 Regression] gcc.c-torture/execute/20051215-1.c is miscompiled with -fgraphite-identity
--- Comment #12 from spop at gcc dot gnu dot org 2010-03-08 17:53 --- Fixed. -- spop at gcc dot gnu dot org changed: What|Removed |Added Status|ASSIGNED|RESOLVED Resolution||FIXED http://gcc.gnu.org/bugzilla/show_bug.cgi?id=43065
[Bug tree-optimization/43065] [4.5 Regression] gcc.c-torture/execute/20051215-1.c is miscompiled with -fgraphite-identity
--- Comment #4 from spop at gcc dot gnu dot org 2010-03-05 22:58 ---
Hi Reza,
In the Graphite branch, we insert in the scop context the bounds on the
parameter kk due to the type of kk:
(gdb) p debug_scop_context (scop)
context (
# eq p0cst
2 3
1 -1 4294967295
1 1 0
)
and then we end up on generating the following CLAST:
(gdb) p debug_clast_stmt (pc.stmt)
for (scat_1=0;scat_1=999;scat_1++) {
S3(scat_1) ;
if (T_7 = 1) {
S4(scat_1) ;
}
S5(scat_1) ;
}
S7() ;
Your reduced case is fixed in the Graphite branch and still fails in trunk:
I will have to merge this change to trunk and I will add your reduced testcase:
commit 67570f6e8a33ff312098b7cca6e4bc5638cf6f7e
Author: spop s...@138bc75d-0d04-0410-961f-82ee72b054a4
Date: Tue Feb 23 13:31:26 2010 +
Add constraints on the type of parameters to the scop context.
2010-02-23 Sebastian Pop sebastian@amd.com
* graphite-sese-to-poly.c (add_param_constraints): Enabled: remove
early return.
git-svn-id: svn+ssh://gcc.gnu.org/svn/gcc/branches/graph...@157002
138bc75d-0d04-0410-961f-82ee72b054a4
For the original testcase, gcc.c-torture/execute/20051215-1.c
I still see that it is miscompiled in the Graphite branch and in trunk.
Sebastian
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=43065
[Bug tree-optimization/43065] [4.5 Regression] gcc.c-torture/execute/20051215-1.c is miscompiled with -fgraphite-identity
--- Comment #5 from spop at gcc dot gnu dot org 2010-03-05 23:04 --- Subject: Bug 43065 Author: spop Date: Fri Mar 5 23:04:17 2010 New Revision: 157247 URL: http://gcc.gnu.org/viewcvs?root=gccview=revrev=157247 Log: Add testcase from PR43065. 2010-03-04 Sebastian Pop sebastian@amd.com PR middle-end/43065 * gcc.dg/graphite/run-id-3.c: New. Added: branches/graphite/gcc/testsuite/gcc.dg/graphite/run-id-3.c Modified: branches/graphite/gcc/ChangeLog.graphite -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=43065
[Bug tree-optimization/43065] [4.5 Regression] gcc.c-torture/execute/20051215-1.c is miscompiled with -fgraphite-identity
--- Comment #6 from reza dot yazdani at amd dot com 2010-03-05 23:10 --- The main difference between the original test case and the reduced one is the type of test expression. I think in the fix you need to treat the address type similar to an unsigned integer. Reza -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=43065
[Bug tree-optimization/43065] [4.5 Regression] gcc.c-torture/execute/20051215-1.c is miscompiled with -fgraphite-identity
--- Comment #7 from spop at gcc dot gnu dot org 2010-03-05 23:14 ---
It seems like the problem is in the overly conservative implementation
of add_param_constraints, as we do not insert bounds for pointers:
if (!INTEGRAL_TYPE_P (type))
return;
and in the original testcase we have a pointer parameter z:
foo (int x, int y, int *z)
{
...
if (z)
b = d * *z;
...
}
--
spop at gcc dot gnu dot org changed:
What|Removed |Added
AssignedTo|unassigned at gcc dot gnu |spop at gcc dot gnu dot org
|dot org |
Status|NEW |ASSIGNED
Last reconfirmed|2010-03-02 20:31:04 |2010-03-05 23:14:28
date||
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=43065
[Bug tree-optimization/43065] [4.5 Regression] gcc.c-torture/execute/20051215-1.c is miscompiled with -fgraphite-identity
--- Comment #8 from spop at gcc dot gnu dot org 2010-03-05 23:46 --- Patch at http://gcc.gnu.org/ml/gcc-patches/2010-03/msg00259.html -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=43065
[Bug tree-optimization/43065] [4.5 Regression] gcc.c-torture/execute/20051215-1.c is miscompiled with -fgraphite-identity
--- Comment #9 from spop at gcc dot gnu dot org 2010-03-05 23:48 --- Subject: Bug 43065 Author: spop Date: Fri Mar 5 23:48:43 2010 New Revision: 157248 URL: http://gcc.gnu.org/viewcvs?root=gccview=revrev=157248 Log: Fix PR43065: Insert bounds on pointer type parameters. 2010-03-05 Sebastian Pop sebastian@amd.com Reza Yazdani reza.yazd...@amd.com PR middle-end/43065 * graphite-sese-to-poly.c (add_param_constraints): Insert bounds on pointer type parameters. * gcc.dg/graphite/run-id-4.c: New. Added: branches/graphite/gcc/testsuite/gcc.dg/graphite/run-id-4.c Modified: branches/graphite/gcc/ChangeLog.graphite branches/graphite/gcc/graphite-sese-to-poly.c -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=43065
[Bug tree-optimization/43065] [4.5 Regression] gcc.c-torture/execute/20051215-1.c is miscompiled with -fgraphite-identity
--- Comment #3 from reza dot yazdani at amd dot com 2010-03-04 18:37
---
I reduced the test to:
--
extern void abort (void);
__attribute__ ((noinline)) int
foo (int *zzz, unsigned int kk)
{
int a, b, d;
a = b = 0;
for (d = 0; d 1000; d++)
{
if (kk != 0)
b = *zzz;
}
return b;
}
int
main (void)
{
if (foo (0, 0) != 0)
abort();
return 0;
}
---
gcc tries to move an invariant unsigned test out of the loop. The following
three guarded loops are generated in the function insert_guard(..) in the
file clast.c.
if (T_7 = -1) {
for (scat_1=0;scat_1=999;scat_1++) {
S3(scat_1) ;
S4(scat_1) ;
S5(scat_1) ;
}
}
if (T_7 = 1) {
for (scat_1=0;scat_1=999;scat_1++) {
S3(scat_1) ;
S4(scat_1) ;
S5(scat_1) ;
}
}
if (T_7 == 0) {
for (scat_1=0;scat_1=999;scat_1++) {
S3(scat_1) ;
S5(scat_1) ;
}
}
S7() ;
$49 = void
kk is an unsigned int in the test program in the guard expression (kk != 0).
The T_7 represents kk in the above expressions. As kk is unsigned T_7 = -1 is
always true and the dereference of the pointer (*zzz) is executed
unconditionally which causes the seg. fault. If the type of kk is changed to
signed, the correct code is generated.
I think, as unsigned expressions are bounded from one side, there must be a
constrain in insert_guard(..) that if the guard is unsigned, only the last two
cases should be generated.
--
reza dot yazdani at amd dot com changed:
What|Removed |Added
CC||reza dot yazdani at amd dot
||com
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=43065
[Bug tree-optimization/43065] [4.5 Regression] gcc.c-torture/execute/20051215-1.c is miscompiled with -fgraphite-identity
--- Comment #2 from pinskia at gcc dot gnu dot org 2010-03-02 20:31 --- Confirmed, the testcase is gcc.c-torture/execute/20051215-1.c by the way. -- pinskia at gcc dot gnu dot org changed: What|Removed |Added Status|UNCONFIRMED |NEW Ever Confirmed|0 |1 GCC host triplet|x86_64-pc-linux-gnu | Last reconfirmed|-00-00 00:00:00 |2010-03-02 20:31:04 date|| http://gcc.gnu.org/bugzilla/show_bug.cgi?id=43065
[Bug tree-optimization/43065] [4.5 Regression] gcc.c-torture/execute/20051215-1.c is miscompiled with -fgraphite-identity
-- mmitchel at gcc dot gnu dot org changed: What|Removed |Added Priority|P3 |P1 http://gcc.gnu.org/bugzilla/show_bug.cgi?id=43065
[Bug tree-optimization/43065] [4.5 Regression] gcc.c-torture/execute/20051215-1.c is miscompiled with -fgraphite-identity
--- Comment #1 from mmitchel at gcc dot gnu dot org 2010-02-17 16:58 --- I take the P1 setting back; I failed to recognize the use of -fgraphite-identity. As long as that's not a default setting, I don't think problems there should be P1. -- mmitchel at gcc dot gnu dot org changed: What|Removed |Added Priority|P1 |P2 http://gcc.gnu.org/bugzilla/show_bug.cgi?id=43065
[Bug tree-optimization/43065] [4.5 Regression] gcc.c-torture/execute/20051215-1.c is miscompiled with -fgraphite-identity
-- rguenth at gcc dot gnu dot org changed: What|Removed |Added Keywords||wrong-code Known to work||4.4.3 Summary|gcc.c- |[4.5 Regression] gcc.c- |torture/execute/20051215-1.c|torture/execute/20051215-1.c |is miscompiled with - |is miscompiled with - |fgraphite-identity |fgraphite-identity Target Milestone|--- |4.5.0 http://gcc.gnu.org/bugzilla/show_bug.cgi?id=43065
